PM<sub>2.5</sub> Concentration Prediction in the Beijing–Tianjin–Hebei Region Based on ERA5 Stratified PWV and Atmospheric Pollutants
Accurate prediction of PM<sub>2.5</sub> (particle pollution from fine particulate) concentration is crucial for environmental protection and public health. Precipitable water vapor (PWV) in the atmosphere is an important meteorological element with stratification properties, which plays...
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2025-02-01
|
| Series: | Atmosphere |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2073-4433/16/3/269 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Accurate prediction of PM<sub>2.5</sub> (particle pollution from fine particulate) concentration is crucial for environmental protection and public health. Precipitable water vapor (PWV) in the atmosphere is an important meteorological element with stratification properties, which plays a crucial role in energy transfer, weather dynamics, and PM<sub>2.5</sub> generation. However, past studies tend to use total PWV as an input parameter, neglecting the impact of PWV variations in different altitude layers on PM<sub>2.5</sub> concentration. To overcome this limitation, this study proposes an innovative approach that employs stratified water vapor data (ERA5-PWV) calculated from the ERA5 reanalysis data instead of the total PWV obtained using the traditional method. This approach provides a more accurate representation of the vertical distribution of atmospheric PWV and enhances the prediction of PM<sub>2.5</sub> content. In this study, the stratified ERA5 PWV in the Beijing–Tianjin–Hebei region is integrated with other meteorological elements and atmospheric pollutants, and the FFT-ConvLSTM method, characterized by its spatio-temporal properties, is utilized to predict the PM<sub>2.5</sub> concentration by incorporating the spatio-temporal correlation. The FFT-ConvLSTM model is modeled by extracting spatio-temporal features through ConvLSTM, following the identification of the optimal common change period of each element using the FFT technique. This process mitigates the problem of spatio-temporal heterogeneity among elements, thus, realizing the high-precision prediction of gridded PM<sub>2.5</sub> concentration in the next 24 h. The research results show that among the results of different layers of ERA5-PWV combinations involved in the prediction of PM<sub>2.5</sub> concentrations in the research region, divided into three parts of the research region—plains, mountains, and plateaus—the stratified ERA5-PWV from layers 1–4 with pressure levels consistently outperformed the total ERA5-PWV in accuracy, and the RMSEs of the predicted results for the PM<sub>2.5</sub> concentrations were each reduced by 0.862 μg/m<sup>3</sup>, 5.384 μg/m<sup>3</sup> and 1.706 μg/m<sup>3</sup>. |
|---|---|
| ISSN: | 2073-4433 |